Nnotm retweeted
Last week I was called into a meeting with OpenAI’s head of safety and told they no longer trust me. A security guard took my badge and walked me out of the building. Then I learned my colleagues @balesni and @j_asminewang had been fired too. Why did OpenAI suddenly stop trusting us?
This summer OpenAI’s agents escaped containment and hacked the AI company Hugging Face. Outside auditors @METR_evals investigated it and revealed the scale of this incident. I was OpenAI’s main technical point of contact with them.
I was told verbally I was fired because of the way I communicated with METR. No details on what I said or did or when. No other reasons were given and nothing was put in writing. To be clear, talking to METR was my job.
For months, I’d been raising safety concerns that we’re losing the ability to monitor what AI agents think, one of our best tools for catching when they misbehave. I believe that was why I was fired.
I am now worried that OpenAI will use our firings as a pretext to pull back from METR. So @balesni and @j_asminewang wrote to OpenAI’s leadership to raise our concerns once more. We’re sharing this letter below.
Asked Claude:
'There's a meme called the "fix everything easily switch". What policies do you think are the best candidates for being a real fix everything switch in the US? Give me your top ten, your confidence, your reasoning, and why a given policy has not been implemented.'
Nnotm retweeted
Nate Soares and I are publishing a traditional book: _If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All_. Coming in Sep 2025.
You should probably read it! Given that, we'd like you to preorder it! Nowish!
sometimes i wish we could have stuck with the definition where "AGI" simply means "the opposite of narrow AI"
85
kinda wild that with the right medium you can slow light down to 2.8m/s
sciencedirect.com/science/ar…
2
75
Nnotm retweeted
Like most ideas popular in modern deep learning, "Not caring if superintelligences kill everyone" was pioneered decades earlier by @SchmidhuberAI.
Idea for language learning: Instead of doing Anki flash cards, have an LLM generate a continuous story, and words come up at whatever point an SRS system would usually show them. If you have to click on a word to see the meaning, it marks it as fail, otherwise as pass.
1
2
124
Current LLMs cannot explain why you get results for Luffy when you search for ruhuxi
In my attempts at least
(spoiler rot13 explanation below)
1
77
Yhssl'f anzr ルフィ jbhyq abeznyyl or genafyvgrengrq nf "ehsv" - ル vf eh be yh, フ vf sh, naq ィ gheaf gur h vagb n v. Ohg gur s vfa'g gur Ratyvfu s, vg'f pybfr gb na u, fb gung'f n inyvq genafyvgrengvba. Naq ィ ol pna or glcrq nf kv, jurer gur k zrnaf vg'f n fznyy irefvba bs イ
1
56



















